Mechanisms of Swelling by Compacted Clay CHARLES C. LADD, Soil Engineering Division, Department of Civil and Sanitary Engineering, Massachusetts Institute of Technology

The mechanisms believed to cause swelling in saturated clay-water systems are f i r s t reviewed. These concepts, drawn primarily from soil chem­istry and soil physics, are extended to compacted natural clays. Test data are presented to show the effects of the ion concentration in the pore fluid on the swelling behavior of a highly plastic clay. These data consist primarily of heaving and fluid pickup measurements on samples molded with pure water and soaked in solutions of varying salt con­centrations .

For the clay investigated, i t is concluded: 1. For samples compacted wet of optimum

water content, swelling can be explained by osmotic repulsive pressures arising from the difference in ion concentration in the double-layer water between interacting clay particles and that in the free pore water.

2. For samples compacted dry of optimum water content, swelling is influenced by factors in additions to osmotic pressures. These other factors may be: the effect of the negative electric and London van der Waals force fields on water, cation hydration and the attraction of the particle surface for water, elastic rebound of particles, a flocculated particle orientation, and the pres­ence of air. The relative importance of these other factors is not known.

• IN SEVERAL areas of the world the differential heaving of foundations due to the swelling of highly plastic clays has resulted in severe damage to buildings. Swelling of clay also changes the engineering properties of strength, compressibility, and per­meability. For many earth structures, such as road subgrades and embankments, a loss of strength or an increase in compressibility wi l l be of greater concern to the soil engineer than the heaving per se. As the use of clay, and especially highly plastic clay, for earth structures increases, the need for a better understanding of the swelling phe­nomenon becomes greater, since the soil engineer must not only know the "as-compacted" properties of clay, but also know how these properties change with time.

The swelling behavior of a compacted clay wi l l be governed primarily (but not solely) by the following factors:

1. Composition of the clay—composition and amount of clay minerals, nature and amount of exchangeable cations, proportions of sand and silt in the clay, and presence of organic matter and cementing agents.

2. Compaction conditions—molded water content, dry density, degree of saturation, and type of compaction.

3. Chemical properties of the pore fluid—both that during compaction and that which is imbibed during swelling.

4. Confining pressure applied during swelling. 5. Time allowed for swelling. This paper is an effort to explain the swelling phenomenon in compacted clays. This

10

11

work is part of a fundamental study of the behavior of fine-grained soils that is cur­rently being carried on by the Soil Engineering Division at MIT.

Data are presented to show the effects of salt content in the pore fluid on the swelling characteristics of a compacted natural clay. The data consist primarily of heaving and fluid pick-up measurements on samples molded with pure water and then soaked in solu­tions of varying salt concentration. The idea for this study came from research in soil physics and soil chemistry by such workers as Bolt (3) and HemwallandLow(^). These workers ran consolidation-rebound tests on fractionated samples of pure clay minerals and showed that an osmotic repulsive pressure is developed between clay particles that causes rebound or swelling when the effective stress (total pressure minus pore pres­sure) on the sample is reduced. This repulsive pressure is proportional to the differ­ence in salt content in the water between the clay particles and that in the "free" pore water.

The author's data indicate that osmotic repulsive pressures play an important role in the swelling of a compacted natural clay since an increase in the salt content of the soaking solution was found to reduce the amount of swelling by the soil. It was found, however, that swelling is influenced by factors in addition to osmotic pressures. Other possible swelling mechanisms are discussed.

Since principles of physical and colloidal chemistry and related fields are important to an understanding of swelling, the f i rs t portion of the paper summarizes those physio-chemical properties of saturated clay-water systems which are thought to influence swelling. These concepts are then extended to compacted natural clays. Only inorganic clays are considered. While the theoretical considerations are somewhat oversimpli­fied and the experimental data and conclusions admittedly incomplete, i t is hoped that the reader may at least gain a better insight into the causes of swelling by compacted clay.

THEORETICAL CONSIDERATIONS Swelling in Saturated Clays

Swelling of saturated clay is considered f irs t , since its behavior is somewhat sim­pler and better understood than that for a partially saturated soil such as compacted clay. An excellent example of the swelling behavior of the former is the rebound por­tion of a standard consolidation test on a saturated clay for which any volume increase in the clay sample is accompanied by an equal (neglecting changes in the density of water) increase in the volume of water in the clay.

The Clay Micelle. — Figure 1 represents schematically a clay particle immersed in pure water. Sufficient exchangeable cations

Boundary of (the effect of the hydrogen and hydroxide ions Double Layer in water are not considered in this paper) sur-

round the particle and are attracted to i t by a net negative charge in the clay particle in or-

© ~v^^ der that the cations plus the particle constitute ® \ an electrically neutral system. This system

© © \ is designated the clay "micelle." The ions ' andwater within the micelle constitute the

© ® ) "double layer." If the clay particle were im-j mersed in a salt solution instead of pure water,

@ ® / then anions would also be present in the double ® /' layer, but the number of cations would be in-

© / creasedaccordingly in order that the micelle / st i l l remain electrically neutral. In other

© / ' words, the double layer includes that portion ^- -—. ^ - ' of the water surrounding the particle in which

there is a negative electric field requiring an excess of positive charges relative to nega-

Figure 1. The clay micelle in pure water. tivecharges.

/ © / ©

V © ©

© ©

\ ® ©

12

A A

-2d-

5- = R - A

The variation in cation and anion con­centration with distance f r o m the clay par­t icle surface fo r suspensions of clay in water can be calculated f r o m principles of colloidal chemistry, the Gouy-Chapman theory being an example (8). Calculations based on such theories are quantitatively correct fo r the ideal systems f o r which the theories were developed; however, most natural soils are fa r f r o m these ideal systems so that one can only use such theories in a qualitative manner when dealing with soils.

Forces Between Clay Particles. — The magnitude of reboimd in a standard con­solidation test i s , of course, directly related to the decrease in effective stress. In turn, the effective stress "a can be related to physicochemical forces acting between clay particles, and i t is these latter forces that are basically responsible f o r swelling. Lambe ( U ) represents this relationship between forces by the expression (Fig. 2)

= R - A

for paral lel particles at an equilibrium interparticle spacing of 2d (this relationship and the following discussion of the forces involved apply to interparticle spacings of

greater than 10 to 20 A . ) where: (assum-

Figure 2. Forces between particles.

two clay

Pg = Osmotic Pressure

Piston

Solution /such as sugarv V in water I \

Pure Water

Semipermeable Membrane

Figure 3. The osmotic solution.

pressure of a

ing unit area)

1. A = the attractive pressure which is usually considered to be caused by London van der Waals secondary valence forces of attraction between the adjacent clay p a r t i ­cles. (This secondary valence force is p r imar i ly a function of particle thickness and interparticle spacing and decreases rapidly with increasing interparticle spacing (8).)

2. R = the repulsive pressure which arises f r o m the interaction of the double layers associated with the two clay p a r t i ­cles. This pressure w i l l be discussed.

A parallel particle alignment is assumed in order to s impl i fy the concept of inter­

particle forces. "Edge effects" arising f r o m electrical charges at the edges of the particle are neglected. These edge effects may cause an additional attractive force between particles that can lead to a nonparallel particle orientation ( n ) .

A l l natural clays w i l l swell or rebound when the effective stress W is reduced and many, part icularly i f remolded (remolding tends to align clay particles into a more nearly paral le l orientation and also tends to break attractive bonds that may have existed between particles at points of contact pr ior to remolding, Lambe (11)), w i l l even slake completely i f a is reduced to zero (for example, the result of immersing an tmsupported chimk of clay in water). Hence, f o r these clays R must be greater than A f o r at least some of the particles, and the mechanism of swelling can be studied by investigating the nature of the repulsive pressure R.

I t is generally believed. Low and Deming (1^), that this repulsive pressure has sev­eral components. One of the most important components is thought to be caused by an osmotic pressure (This osmotic pressure is the "electr ic l repulsion" re fe r red to by other wr i te rs such as Lambe (10, U ) . The higher ion concentration causing osmotic pressures arises, after a l l , f r o m the electric f i e l d in the double layer . ) arising f r o m

13

/ ©®

- Doubia Loywa Overlap

-Imogmory SainipermaabI* Membrane Surrounding

Cloy Rarticlee

Figure k. Osmotic pressure between clay particles immersed in water.

two

the higher ion concentration in the double-layer water of the clay micelle than in the "f ree" pore water, that is, that water out­side of the micelle. The nature of the other components w i l l be presented after a discussion of osmotic pressures.

Osmotic Repulsive Pressure (Pr inc i ­ple) . —When an aqueous solution is separated f r o m pure water by a semi-permeable membrane, that is, a membrane that per­mits the passage of water but not that of the substance dissolved (solute) in the so­lution, water tends to pass through the membrane into the solution, thereby d i lu ­ting i t . This phenomenon is called "osmo­sis. " The pressure that must be applied

Silt Particle

Air-Water Interface

to the solution in order to prevent the flow of water into the solution through the semi­permeable membrane is called the "osmotic pressure" of the solution. This is shown diagramatically in Figure 3.

Although a mechanistic picture has not yet been developed to explain fu l ly osmosis, there are formulas f o r calculating osmotic pressures. One of the simplest of these is the van't Hoff equation which yields

Po = RTc

where_Po = osmotic pressure (g per sq cm), R = gas constant, T = absolute tempera­ture (RT = 2. 5 X lO'' g cm per mole f o r 20 C), and c = concentration of solute (moles per cc of solution). (The van't Hoff equation as presented is only s t r ic t ly applicable to very dilute solutions of nonelectrolytes. Measured osmotic pressure values exceed calculated values f o r most other cases, Prutton and Marion (17).) Osmotic pres­sures can reach very large magnitudes. For instance, 130 gm of sugar per l i t e r of aqueous solution exerts an osmotic pres­sure of about 10 tons per sq f t . Osmotic pressures can, of course, be developed between two solutions of unequal concen­tration so that

Po = RT (ca - Cb)

where c^ and Cb refer to the solute concentrations on either side of a semi-permeable membrane.

Osmotic Repulsive Pressure (in Clays). —Osmotic pressures can act in clays since: (a) there exist differences in solute concentrations (in this case ions are the solute), and (b) the electric f i e l d around the negatively charged clay part icle acts as a semi­permeable membrane. This can be illustrated by picturingtwo clay particles that have been immersed in a beaker of pure water and pushed under an effective stress or to an interparticle spacing of 2d, as shown in Figure 4. (Pure water is used to s impl i fy the i l lustrat ion, because in this case the only ions in the double layer w i l l be the exchange­able cations.) Because of the exchangeable cations, the concentration of ions in the double-layer water is larger than the concentration of ions in the f r ee water. The exchangeable cations are attracted to the clay particles by the negative electric f i e l d arising f r o m the negative charge on the particles. Hence the electric f i e ld acts as a semi-permeable membrane in that i t w i l l allow water to enter the double layer but w i l l not allow the exchangeable cations to leave the double layer. The dashed line in Figure 4 depicts f iguratively this "semi-permeable membrane." One can see that due to the difference in ion concentrations between points such as "a" and "b", water would l ike to f low f r o m "b" to "a" and that an effective stress a (plus the attractive pressure A)

Clay Parti Cluster

Figure 5. An i l l u s t r a t i o n of pore tensions i n compacted cle^.

water

14

is required to prevent an increase in inter-particle spacing, that is , swelling. If a

^ is reduced, then water f lows f r o m "b" to "a", thus decreasing the ion concentration

Capillar Tube o^^^ words, the double layer is (Tscosac ^ a I ary u e expanded) unt i l the correspondingly lower

Water ^ p ^ r A Water ir osmotic pressure is again in equilibrium ^-fa * * wi th the effective and attractive pressures.

At this point one might re-examine the p^—f^(»stx meaning of the t e rm "pore water pressure" where- Ts • surface tension in water as used in soi l engineering. When the soil

r = radius of capillary tube engineer measures pore pressures, as f o r tx' contact angle instance with a piezometer, he measures

the pressure in the "free pore water" (that Figure 6. Air preasure In capil lary tube. water outside of the double layer) . For a

particle arrangement as shown in Figure 4, the pore pressure u as used in soil engineering is , therefore, the pressure in the water as a point such as "b" . The total pressure in the double-layer water (to mean water plus Ions) at a point such as "a" midway between clay particles is , however, greater than u by a pressure equal to the developed osmotic pressure (and possibly other pressures to be mentioned) and this pressure increase is counterbalanced by (a + A) . This assumes that the pore pressure u also acts between particles. For a detailed discussion of var­iations in water pressures throughout a clay-water system, see Low and Deming (12).

Osmotic Repulsive Pressure (Factors Influencing).—The ion concentrationdifferen-tial that determines the osmotic repulsive pressure between particles is the ion con­centration at the midplane between particles (that is, point "a" in Fig . 4) minus the ion concentration in the water outside of the double layer (the f ree pore water). Thus, based on the van't Hoff equation, the osmotic pressure becomes

Po = ^ T ( c ^ - c^) where Cg and C Q refer to the total ion (both cations and anions) concentration (moles per cc) at the midplane and in the f ree pore water, respectively. (The reader is r e ­f e r r e d to Bolt (3), Kruyt (8), Low and Deming (12), Hemwall and Low (5) and Lambe (11), f o r a more detailed presentation and discussion of these and other factors and f o r acIHitional reference material.)

Whereas the ion concentration in the f ree pore water C Q can be easily measured, the midplane concentration c^ must be computed f r o m a theory relating ion concentration with distance f r o m the clay particle surface. For certain ideal cases, the Gouy-Chap-man theory has been used and the calculated osmotic pressures checked experimentally. Bolt (3) andWarkentin, e ta l . (22) ran consolidation-reboimd tests on samples of f r a c -tionafed (minus 0.0002 mm) monlmorillonite and i l l i t e with various exchangeable cations and pore water ion concentrations over a pressure range of f r o m 0 .1 to 50 tons per sq f t . Their data show that the compression-swelling curves based on osmotic pressures computed f r o m the Gouy-Chapman theory agreed qualitatively and in some cases almost exactly wi th the observed curves, and i n a l l cases the observed trends corresponded with theory. (Bolt and Warkentin, et a l . (22) hypothesize that unknown variations in p a r t i ­cle shape and spacing can account fo r much of the discrepancy between the computed and measured curves.)

Of main interest, the Gouy-Chapman theory tells us (as does intuition) that fo r a given clay:

1. For a constant interparticle spacing, PQ decreases with increasing ion concen­trat ion in the f ree pore water.

2. Pg decreases with increasing interparticle spacing.

Furthermore, the data by Bolt and others indicate that variations in osmotic pres­sures, and hence swelling, as predicted by the Gouy-Chapman theory, should also^apply in a qualitative sense to natural clays f o r interparticle spacings exceeding 10 to 20 A. A quantitative treatment of natural clays is , of course, impossible because the extreme

15

variations in particle size, shape, composition, orientation, and spacing preclude any realistic computation of interparticle forces.

Other Repulsive Pressures and Swelling Mechanisms. — The previous discussion has shown that an osmotic pressure can reasonably account for a repulsion between clay particles which results in swelling, or in other words, an expansion of the double-layer, when the effective stress on a clay sample is reduced. Other factors may also contribute to swelling. One of these is the effect of secondary valance or London van der Waals forces on the water surrounding clay particles, Hemwall and Low (5) and Low and Deming (12). Another is the effect of the negative electric f i e ld on the double-layer water. Both of these force fields are believed to attract water to the soi l p a r t i ­cles, although the latter force f i e l d is not thought to contribute to R fo r part icles at equilibrium. Low and Deming (12). Whereas there are f a i r l y extensive data indicating that osmotic pressures play a very important role in swelling, at least fo r certain clay-water systems, there are no known data f r o m which definite conclusions can be drawn relative to the magnitude of influence of the above two factors on swelling.

As previously noted, the foregoing^discussion has been restr icted to interparticle spacings, 2d, greater than 10 to 20 A. (The double-layer thickness would then be 5 to 10 A . ) For smaller spacings, the nature and interrelationship of o", R, and A may change radically. (At these small interparticle spacings, such as one might expect in a highly compressed clay, the likelihood of having A greater than R may be increased fo r many of the interacting particles, Lambe (11), p 1655-20.) For these smaller spacings (among other things):

1. Osmotic pressures may not be developed; in other words, the "normal" double-layer is not formed, MacEwan (13) and Norr i sh (15).

2. Water adsorption is generally thought to be governed p r imar i ly by hydration of the exchangeable cations and the attraction of the clay particle surface fo r polar mole­cules l ike water, Barshad (2), MacKensie (li) and Norr ish (15).

Mention should be made of swelling due to elastic rebound and "imbending" of soil particles. I t is fe l t , Lambe {11), that this phenomenon, while important in coarse­grained soils and in soils containing relatively large platey shaped particles (like mica), is of l i t t le importance fo r most clays. Bolt (3) discusses the importance of these "mech­anical" effects versus the "physicochemical" effects fo r natural soils .

Swelling in Compacted Clays

In attempting to util ize the foregoing swelling concepts to explain the swelling be­havior of a compacted clay, one is confronted with the following complications. As the molded water content changes (the reader is re fe r red to Holtz and Gibbs (7) fo r ex­tensive data on the swelling characteristics of compacted clays) fo r a given compactive effor t :

1. The dry density of the soi l and thickness of the double-layer water around clay particles vary.

2. The particle orientation varies. 3. The pore water tensions vary. 4. The degree of saturation, and hence the amount of a i r in the sample, varies.

The pressure in the air may also vary.

The above factors are discussed relative to their possible effects on swelling be­havior and the mechanisms involved.

Effect of Molded Water Content on Dry Density and Double-Layer Thickness. —The thickness of the double-layer water on compacted clay particles is nearly always less than that which the particles would like to have i f given free access to water. (Silt and sand size particles also have "double-layer water ." However, these particles have a relatively small specific surface area that is , surface area per unit weight of soi l , so that the effect on the over-a l l volume change of anexpansionof the double.layer around these particles is small compared to that f o r clay size particles where the specific sur­face area is large.) For instance, many plastic clays under zero confining pressure

16

w i l l imbibe water unt i l they reach a water content approximating the l iquid l i m i t , whereas the clays may be compacted at water contents less than the plastic l i m i t . The difference be­tween these two water contents has been termed the "double-layer deficiency" Lambe (11). This deficiency is thought to be caused by the same factors causing swelling, that i s , a Bau­ble-layer expansion, in saturated clays (for example, osmotic pressures, effectof force fields on double-layer water, e tc .) . If the thickpess of double-layer water is less than 5 to 10 A (corresponding to a 2d spacing of 10 to 20 A) the double-layer deficiency may in part be due to cation hydration, etc., as previously mentioned. A double-layer thickness of 5 to 10 A is approximated by the water content of the soi l at 99 percent relative humidity. This water content, evenfor extremely plastic soils, seldom reaches 10 to 15 per cent (pure mont-moril lonite is an exception) and is usually only a few percent f o r most lean clays. Hence fo r many clays, the water content used for compaction may be sufficient fo r a double-layer thick­ness of greater than 5 to 10 A.

The thickness of the double-layer is roughly proportional to the molded water content (assuming a l l water to be in the double-layer, the average thickness of the double-layer is then equal to the water content divided by the specific surface area). Hence the lower the molded water content (other things being equal), the greater is the water uptake required to satisfy the double-layer deficiency; andfor a constant molded water content, an increase in dry density would lead to an increase in the amount of swelling.

Effects of Molded Water Content on Particle Orientation. —There is evidence, Pacey (16), to show that the orientation of clay particles changes with molded water content. These data indicate that compaction dry of optimum water content leads to a nonparallel or f loccu­lated orientation while compaction wet of optimum leads to a paral lel or dispersed orienta­tion of clay particles. Different methods of compaction may also yield different particle orientations, even at the same density and water content. One might expect different pa r t i ­cle orientations to cause a difference in swelling behavior. Seed's (18) data suggest that a flocculated particle orientation w i l l swell more than a dispersed orientation.

Effects of Molded Water Content on Pore Water Tensions and Degree of Saturation. — Pore water tensions (water pressures less than atmospheric) imdoubtedly exist in compacted clay, part icularly i f compacted dry of optimum water content where the in i t i a l degree of sat­uration is we l l below 100 percent, Aitchison (1.), Soil Mechanics f o r Road Engineers (19) and Hil f (6). Tensions are caused by the double-layer deficiency in the clay micelles. (As pre­viously pointed out, one should also include the larger size particles since they also desire water . ) In other words, a l l the particles in the soi l are competing f o r the l imi ted supply of water in order to expand their double-layers. Capillari ty may also enter the picture i f the voids in the soi l contain both air and "f ree" water. * The desire of the clay micelles to imbibe this f ree water would be resisted by the surface tension at the air-water interface in the void.* An i l lustrat ion of this phenomenon is shown in Figure 5 which depicts clusters of clay p a r t i ­cles compacted between s i l t particles. The inflow of water to the clay is restrained by the menisci in the pores between the s i l t particles. This i l lustrat ion is only one example of the many possible situations that might occur, since a compacted clay w i l l contain a wide var ia­tion in the size and shape of particles and voids.

In summary, then, pore water tensions in a compacted clay represent a balance between double-layer deficiencies and surface tensions at air-water boundaries. Another way one might look at pore water tensions is to say that they exert an effective stress

* Capillari ty in the same sense as the capillary r ise of water in a glass tube. The author, when using this te rm, is re fe r r ing in general to relatively large pores in the soi l (for example, diameters of several thousand A or larger) where the amount of double-layer water is relatively smal l compared to the "f ree" pore water. Although the fundamental cause of capillari ty may originate f r o m the electrical nature of the soi l particles, the t e rm capillari ty is a widely used concept which seems applicable in this case, a

Calculations based on the capillary equation, Taylor (20), relating pressure, surface tension, and pore diameter show that a 3,000 A diameter pore can resist a tension of about 10 atmospheres. Bolt 's (3) data f o r Na montmorillonite show that a swelling pressure of 10 atmospheres corresponds to a particle spacing of only about 30 A (dou­ble-layer thickness of 15 A ) .

17

on the compacted clay mass which prevents the particles f r o m imbibing water and swelling. This effective stress is more of an "effect" than a "cause."

When a compacted clay sample is put in contact wi th water, any air-water menisci at the surface of the sample are broken. Water w i l l flow into the clay because of the pore water tensions within the sample, these tensions having been caused by a combi­nation of double-layer deficiencies and capi l lar i ty . With t ime, the pressure in the pore water increases to atmospheric with a resultant lowering of the effective stress within the sample. Concurrently, the clay micelles expand their double-layers and swelling occurs between clay particles, just as fo r the saturated clays, unt i l the repulsive pres­sure minus the attractive pressure between particles is in equilibrium with any applied effective stress. Over-al l swelling of the sample can, of course, be prevented by ap­plying an effective stress to the soi l equal to the "swelling pressure."

The Role of A i r in Swelling. —Previous discussion has mentioned that the presence of a i r can influence the magnitude of the pore water tensions that are developed in com­pacted clay. A i r may also influence the swelling behavior in another manner. Swelling data, Holtz and Gibbs (7), Figures 8, 10 and 11, usually show that the total volume of a i r in a compacted sample decreases during the soaking process, part icularly fo r sam­ples compacted dry of optimum water content, although the f ina l degree of saturation is s t i l l less than 100 percent. Thus, during the soaking process, some of the a i r i n i ­t ia l ly in the soil voids must either escape f r o m the soi l , be dissolved by water, or be compressed by capillary forces. Most l ikely a combination of these conditions occurs. If a nonspherical pocket of a i r is compressed in a soil void during the soaking process, the pressure in the air may produce tensile stresses in the soi l skeleton forming the void in which the air resides (̂ a long, ini t ia l ly a i r f i l l e d pore having a relatively large diameter of several thousand A or larger is visualized). These tensile stresses could cause an increase in the volume of the void, and hence an expansion of the soi l .

An analogous situation is shown in Figure 6. An ini t ia l ly a i r f i l l e d capillary tube is immersed horizontally in water. As water enters the tube f r o m both ends by capil lari ty, the air in the tube becomes compressed. An analysis of the stresses on a cross-section of the tube through the air pocket (there must be enough a i r in the tube so that the air pocket does not become spherical) shows that both axial and hoop tensions act in the tube at this location; in other words, the pressure in the air tends to expand the tube. If the capillary tube were br i t t le and could not withstand these tensions, i t would ultimately break at the center. If the capillary tube were f lexible at the center, a bulging would occiu-. The maximum air pressure that can be attained in a capillary tube of a given radius (assuming a sufficient amount of a i r ) is proportional to the tensile strength of the tube or the surface tension of water, whichever is smaller .

In a s imilar manner, i t is believed that a i r in a soi l void, if under pressure, can cause an enlargement of the void i f the soi l skeleton cannot resist this pressure. These air pressures could be quite large. For example, theoretically, a i r could be compres­sed by capillary forces to a pressure of 6 atmospheres in a void having a diameter of 0. 5 microns (5, 000 A) . Such a pore size is not unlikely in a compacted clay.

One can probably best visualize this swelling phenomenon occurring in a soi l where: (a) There are many interconnected tubular a i r voids; (b) the air pressure is ini t ia l ly atmospheric, and (c) water enters the soil f r o m a l l directions during soaking. These conditions (except possibly f o r the shape of the voids) are met when a dry sample of clay is immersed without confinement in water. The rapid slaking of such a sample is said, Terzaghi and Peck (21), p . 129, to be caused at least in part, by the a i r pres­sure that is built up within the sample. Data w i l l be presented which suggest that a i r pressures may contribute to the swelling of clay i f compacted dry of optimum water content.

The above discussion dealt wi th the presence of a i r in the larger voids of a com­pacted clay. The presence of a i r in the double-layer water between interacting clay particles would also affect swelling. For example, i f a i r were present, i t would tend to expand the double layer, since i t is displacing water and ions. In other words, fo r a given interparticle spacing, the presence of a i r would increase the repulsive pres­sure; or f o r a given effective stress, the interparticle spacing would be larger. The magnitude of influence on swelling of this a i r is , however, not known. Even the amount

18

of a i r to be expected in the double layer is a matter of conjecture. Only two "types" of a i r have been discussed, that is , large, f a i r l y continuous voids

of a i r that extend throughout the clay mass, and minute bubbles of a i r that exist in the double-layer water. While air imdoubtedly exists in many forms in compacted clay, the above conditions are thought to represent two extreme cases which can be used fo r a consideration of the major effects of a i r on the swelling behavior of compacted clay.

EXPERIMENTAL DATA

Swelling data are presented in which compacted samples of a clay are soaked in aqueous solutions ranging f r o m pure water to a 5 molar salt solution. According to the previously developed theory, the samples which are soaked in salt solutions should swell less than the samples soaked in pure water, since an increase in ion concentration in the water outside of the double layer reduces the osmotic repulsive pressure between clay particles. In other words, the salt solution w i l l reduce the double-layer deficiency in the clay micelles. Data are also presented relative to the distribution and pressure of a i r in as-compacted samples of clay.

Description of Soil

The soi l , Vicksburg Buckshot clay, was supplied to the M I T Soil Stabilization Labo­ratory by the Waterways Experiment Station, Vicksburg, Mississippi . The Atterberg l imi t s , specific gravity, grain size distribution, mineralogical composition, and cer­tain other properties are presented in Table 1. Of particular interest are: (a) The highly plastic characteristics of the soi l , although only 36 percent of the soi l is clay size; (b) the presence of montmorillonite, a clay mineral known to be very expansive;

TABLE 1

PROPERTIES OF VICKSBURG BUCKSHOT CLAY

1. Specific gravity^ = 2.74

2. Atterberg l imi t s^ Liquid l i m i t = 63 percent Plastic l i m i t = 25 percent Plasticity index = 38 percent

3. Grain size distribution 94 percent minus 0.074 mm; 36 percent minus 0.002 mm

4. Mineralogical composition in percent by weight^

Montmorillonite } ^ t e r s t r a t i f i ed Quartz 20-3 Feldspar 20 - 10 FezOs 1 . 1 ^ 0 . 1 Organic matter 1 . 1 ^ 0 . 1

5. Other properties' ' Soluble salts (meg. NaCl/100 g) = 0.3 Cation exchange capacity (meg./lOOg) = 30; 52 (minus 0.002 mm) Glycol retention (mg/g) = 65; 135 (minus 0.002 mm) pH = 4.9

Note: Items 4 and 5 obtained by R. T. Mar t in , Research Associate, MIT Soil Engineering Division.

^ On a i r -d r i ed and ground so i l . Unless otherwise specified, data shown f o r soi l passing 0.074 mm. t indicates probable uncertainty in percentages given.

19

and (c) the high glycol retention by the so i l . Multiplying the glycol retention by 3.22 to obtain the specific surface area yields values of 210 and 435 sq m per g of clay, r e ­spectively, f o r the two soi l fractions (by comparison Na montmorillonite has a surface area of 800 sq m per g). The exchangeable cations are believed to be predominantly calcium.

Test Procedure

The test procedure employed fo r the compaction-soaking tests was as follows:

1. The clay was a i r -dr ied , ground to pass a No. 20 sieve, and equilibrated fo r at least two days at the desired water content. Dis t i l led water was used as the molding f l u i d .

2. The clay was compacted dynamically in 2.75-in. diameter by 0.85-in. consoli-dometer rings (fixed r ing type) at an effor t of 28,000 f t - l b per cu f t and t r immed to a height of 0.6 i n . A surcharge of 200 lb per sq f t was applied, followed by immersion of the sample in an aqueous solution. The test set-up is shown schematically in F ig ­ure 7. Note that the soaking solution enters both ends of the compacted sample.

3. Three different aqueous solutions were used fo r soaking. These were: (a) Dis­t i l l ed water; (b) a 0. 5 molar CaClz solution (55 g CaCU per 1,000 cc of solution. CaCla was used to prevent ion exchange); and (c) a 5.0 molar CaCl2 solution. The solubility of CaCls in water at room temperature is approximately 5.2 molar. Hence, a 5 molar CaCl2 solution is very strong.

4. Measurements of the amount of heave were taken over a four day period. 5. At the end of four days the weight of the sample was measured. The weight of

solution in the sample was divided by the estimated unit weight of the solution to obtain a volume of solution. The volume of solution was treated as i f i t were pure water f o r computing water contents after soaking. (That is , i f 10 g of in i t ia l ly dry soi l imbibed 1.5 g of a salt solution wi th a unit weight of 1. 5 g per cc, the recorded "water content" change would be 10 percent. For in i t ia l ly moist samples, however, the exact unit weight of solution in the sample after soaking was unknown and had to be estimated.)

TABLE 2

WATER ADSORPTION AND IMBIBITION DATA ON VICKSBURG BUCKSHOT CLAY

A. Water vapor adsorption^

Relative Humidity (%) Equil ibrium Water Content (%)

50 6 ~ 9 9 13.5

B. Water imbibition under "Free-Swell" condition''

In i t ia l Water Content (%) Equil ibrium Water Content (%)

3.6 64 4.7 68 4.7 67 4.7 60 8.2 55

32.4 53 50.2 51 51.0 54

* On loose samples of in i t ia l ly oven-dried clay. ' ' O n 1.5 mm thick samples of clay spread over a porous stone whose top surface was slightly above a water surface which was enclosed in a sealed container. Equil ibrium content taken after 24 hr .

20

200 lb per sq. ft

Porous Stone (too a bottom)

Figure 7. Compaction-soaking test ratus.

appa-

Swelling Data

Data on water adsorption and water imbibition under a " f ree-swel l" condition are presented in Table 2. The data indicate that: (a) At water contents in excess of 13. 5 percent (the equilibrium water content at 99 percent relative humidity), i t i s reasonable to assume that the exchangeable cations are hydrated and that the normal double layer is formed; and (b) under zero confining pressure, the double layer w i l l imbibe approxi­mately 60 percent water. (Assuming that a l l the water is associated with the clay-size fract ion of the soi l and that the surface area of this fract ion is 435 sq m per g, the average thickness of water around the clay particles in A is approximately one-half of the water content in percent; f o r example, a water content of 60 percent corresponds roughly to an average water thickness of 30 A . ) The molded water contents used f o r the compaction soaking data presented below range f r o m 14 to 24 percent with optimum water content at 19 percent.

The effects of salt concentration on swelling are shown in Figure 8 in which the molded dry density, the percent heave (change in height divided by in i t i a l height times 100), and the "water content" increase are plotted against the molded water con­tent. The effect of salt content on the rate of swelling fo r samples compacted wet and dry of optimum water content are presented in Figure 9. Dry density and water con­tent curves before and after soaking are plotted in Figure 10 f o r samples immersed in both pure water and in the 5 molar CaCla

solution. Figure 11 shows the relationship between volume change (cc) and water pick­up (g) fo r samples soaked in water (since the swelling data were obtained on soi l samples of unequal volume pr io r to soaking—variations of up to 15 percent—the observed data have been adjusted to correspond to in i t ia l samples volumes of 100 cc).

The data show that:

1. The amount of heaving and water pickup decreases with increasing molded water content.

2. The degree of saturation after soaking is less than 100 percent. The volume of water pickup also exceeds the volume of expansion, part icularly for samples compacted dry of optimum water content.

3. The soaking of compacted samples in salt solutions produces a marked decrease in the amount of f l u i d pickup and heaving. Figure 8 shows that the absolute magnitude of this reduction in f l u i d pickup and heaving is f a i r l y uniform, part icular ly f o r the strongest salt solution, over most of the molded water content range. Furthermore, the 5 molar salt solution prevents swelling fo r a sample compacted 2 percent wet of optimum water content.

4. The in i t ia l rate of swelling is practically unaffected by the salt concentration in the soaking solution. (A 5 molar CaCU solution has a viscosity 10 times and a surface tension 1.3 times that of pure water so that one might expect a net decrease in the rate of swelling due to the large increase in viscosi ty.)

The data show that the salt content in the pore f lu id has a decided effect on the swel l ­ing behavior of this compacted clay. Hence i t wouldappear that osmotic repulsive pres­sures play an important role in swelling, since osmotic pressures depend upon the d i f ­ference in ion concentration in the water between the clay particles and that in the f ree pore water. Furthermore, i t would seem that the osmotic pressure concept can satis­factor i ly explain, at least in a qualitative sense, the observed swelling of samples com­pacted wet of optimum water content, since the addition of salt prevented swelling. This does not necessarily mean, however, that an osmotic pressure is the only com­ponent of the repulsive pressure R. For example, even f o r the sample that did not swell when immersed in the 5 molar CaCl2 solution, there must be a repulsive pressure

21

between particles in order to counterbalance the effective stress ( that i s , the 200 lb per sq f t surcharge) plus any attractive stresses that may act. If this strong salt so­lution reduced osmotic pressures to zero, then some other mechanism of repulsion must be operative even though the magnitude of its influence is smal l .

While i t is evident that some of the swelling of samples compacted dry of optimum water content can be explained by osmotic pressures, there are certainly other factors which influence the swelling behavior. One might assume that the reduction in swelling due to the 5 molar CaCl2 solution is approximately equivalent to the "osmotic" swelling

20

on U J

n LU or (J

X

UJ

I

3 O

>-t to z U J

o >• or o

I 6 |

12 -

8

4+

12

10

8

6

4

24

0 108+

104

lOO

Note: Water contents adjusted for salt solutions

(see text)

Soaked in Symbol water

0.5 Molar CaClg 5 Molar CaCIa

16 18 20 22 MOLDED WATER COIMTEIMT (%)

Figure 8. Effect of s a l t concentration on swelling behavior.

22

^ Compacted at

Symbol Soaked In 0 water • 0 5 Molar CaCIa

5 Molor CaCI:

• • 1

Optimum Plus 2 to 3 %

Compacted of Optrmum Minus 2 to 3 %

•I-20 40 60 TIME (hours)

Figure 9. Effect of s a l t concentration on rate of swelling.

5 106

~ 102

90

Ssolud rn soaking

9 Molar CoCI:

Note WOter Contents Adjusted for Salt Solution

(See Text)

14 16 18 ao 22 24 26 28 30 32 WATER CONTENT (%)

Figure 10. Effect of s a l t concentration on density and water content after

soaking.

that occurs when the samples are soaked in pure water. However, an increase in salt concentration may also influence other interparticle forces besides osmotic pressures. A better assumption would be that the 5 molar salt solution reduces swelling due to os­motic pressures to a negligible amount, since rough calculations show that the concen­trat ion of ions in a 5 molar CaCla solution f a r exceeds the concentration of the exchange­able cations between clay particles. (For a monomolecular thickness of water on the

16+ Relationship computed for sample volumes equal to

100 CO before soaking

Wet of Optimum

Dry of Optimum

12 16 20

WATER PICKUP (gm)

24 28

Figure 11. Relationship between volume change and water pickup for samples soaked in water.

23

clay, the average concentration of exchangeable cations, assumed to be calcium, around the particles is only about 2 molar. Reducing the surface area by a factor of two s t i l l only yields a 4 molar concentration. Hence the total ion concentration in a 5 molar CaCl2 solution fa r exceeds the concentration of exchangeable cations in the soil for a l l water contents.)

50

c E u u

<

40+

30+

20+

10+

114 +

I n o X)

5_ 106

to g '02 o >• 98 a

Symbol Compactive E f f o r t 5 Layers 25 Blows 80 lb Tamp 10 "

Compaction in Harvard Miniature size permeameter

cylinders

Pressure Head = 5 lb per sq. in.

> t t — o — A - o -

•+- •+- - I -8 10 12 14 16 18 20 22

MOLDED WATER CONTENT (%) 24

Figure 12. Air flow data.

24

The increased swelling dry of optimum over and above that which might be attributed to osmotic pressures may be due to a number of factors. As previously mentioned under Theoretical Considerations, there are mechanisms other than osmotic pressures by which clay particles can imbibe water. These were: (a) The effect of the negative electric and London van der Waals force fields on water; and (b) the effect of cation hydration and the attraction of the particle surface fo r water, at least f o r small double-layer thicknesses. The relative importance of the f i r s t factor i s unknown; the second factor should be relatively important fo r the molded water content range investigated, since the molded water contents exceeded the equilibrium water content of the soil at 99 percent relative humidity. Elastic rebound and a flocculated particle orientation may contribute to the increased swelling dry of optimum; the magnitude of their i n ­fluence is not definitely known. Finally, there is the role of a i r in swelling. The i n ­crease in volume of f l u i d pickup with decreasing molded water content can be part ial ly explained by the decreasing in i t i a l degree of saturation (increasing volume of a i r voids). The following data suggest that a buildup of a i r pressures within soil voids during the soaking process may contribute to the increased swelling dry of optimum water content.

Data Relative to the Role of A i r Figure 11 has shown that the volume of a i r in the compacted clay samples decreases

upon immersion, part icularly dry of optimum water content, since the volume of water pickup exceeded the total volume increase of the samples during soaking. I t seems possible that some of this decrease in the volume of air voids is caused by a compres­sion of a i r by capillary forces. An increase in the air pressure (preliminary attempts to measure the pressure in the air voids after soaking proved unsuccessful) within en­trapped nonspherical pockets of a i r , as previously discussed, is thought to cause swel l ­ing of so i l .

One would expect swelling due to air pressures to be of greatest importance in com­pacted samples containing numerous interconnected air voids p r io r to soaking. With decreasing molded water content dry of optimum, the volume of a i r voids in the com­pacted clay increases. Data are presented in Figure 12 which indicate that at least some of the a i r in the samples compacted dry of optimum is also continuous (and there­fore at atmospheric pressure), whereas the a i r i n samples compacted wet of optimum appears to be discontinuous. Clay samples were compacted at two compactive efforts in a constant head permeability apparatus, an air pressure applied at the top of the sample, and the quantity of a i r f low measured with a rotameter at the outlet. For both compactive efforts the quantity of air f low decreased rapidly with increasing molded water content, and at optimum water content the flow became too small to measure (less than 1 cc per minute).

The data thus show that the samples having the greatest amount of swelling also ini t ia l ly have the greatest amount of interconnected air voids. While this fact supports the a i r pressure hypothesis, the author has no data to prove its validity. Further ex­perimentation is planned.

Reliable data have not been obtained relative to the amount of air present in double-layer water.

CONCLUSIONS

The data have shown that the salt content in the pore f lu id has an important effect on the swelling behavior of Vicksburg Buckshot clay. Furthermore, i t appears that the osmotic pressure concept can satisfactorily explain a good portion of the swelling that occurs when this clay is soaked in water, part icularly f o r samples compacted wet of optimum water content. Dry of optimum water content, however, swelling is i n ­fluenced by factors in addition to osmotic pressures. These other factors may be: the effect of negative electric and London van der Waals force f ields on water, cation hydration and the attraction of the particle surface fo r water molecules, elastic r e ­bound, particle orientation, and the presence of a i r . The relative importance of these other factors is not known.

Although the experimental data are l imi ted to one clay, the theoretical considerations

25

indicate that many of the same concepts can be extended to other compacted soils. With other soils, however, the relative importance of the different swelling mechanisms may be altered.

An understanding of some of the factors which influence swelling may help the so i l engineer predict, at least in a qualitative sense, the swelling behavior of compacted clays, and i t may even enable him to alter swelling behavior to better suit his needs. This is i l lustrated by the following examples:

1. Fat clays, with their relatively large surface area and hence greater proportion of double-layer water, almost always swell more than lean clays.

2. Clays compacted at low water contents, where the water deficiency in the double layer is high and the degree of saturation is low, w i l l often swell more than clays com­pacted at high water contents.

3. The replacement of low valency exchangeable cations by higher valency cations (for example, calcium fo r sodium) can reduce swelling, since the number of exchange­able cations in the double layer is reduced.

4. The mixing of salt wi th a compacted clay can reduce swelling, since the ion con­centration in the pore water is increased. This may part ial ly explain why the treat­ment of road subbases and sul^rades with salt often improves stabili ty.

5. Leaching compacted clays with salt solutions w i l l result in less expansion and hence less strength loss than leaching with pure water. Salt leaching might be employed to increase the stability of a dam or to reduce heaving at the bottom of an excavation.

ACKNOWLEDGMENTS

Most of the test results presented in this paper were obtained by Jean Charron, fo rmer M I T gradxiate student working under the supervision of the author and Dr . T. Wi l l i am Lambe, Head of the M I T Soil Engineering Division.

Dr . Lambe, Professor R. V. Whitman, Dr . R. T. Mar t in , Professor J . E. Roberts, and R. M . Harkness, a l l of the MIT staff in Soil Engineering, cr i t ica l ly reviewed this paper and made many valuable suggestions f o r i ts improvement. Dr . Mar t in was par­t icular ly helpful in explaining the enigmas of colloidal and physical chemistry to the author. The author, however, accepts a l l responsibility fo r the material presented.

The Waterways Experiment Station, Corps of Engineers, U . S. Army, sponsored the thesis work by the author on this subject. Their assistance is gratefully acknowl­edged.

REFERENCES

1. Aitchison, M . E . , "The Strength of Quasi-Saturated and Unsaturated Soils in Re­lation to the Pressure Deficiency in the Pore Water ." Proc. of the Fourth International Soils Conference, Vol . I , p. 135 (1957).

2. Barshad, I . , "Adsorption and Swelling Properties of Clay-Water Systems." Bu l l . 169, Division of Mines, State of California (1955).

3. Bolt, G. H . , "Physico-chemical Analysis of the Compressibility of Pure Clay." Geotechnique, Vol . 6, No. 2, p . 86 (1956).

4. Charron, J . , "Swelling of Compacted Clay ." M.S. Thesis, Mass. Inst, of Technology (1958).

5. Hemwall, J . B . , and Low, P. F . , "The Hydrostatic Repulsive Force in Clay Swelling." Soil Science, Vol . 82, No. 2, p . 135 (1956).

6. Hi l f , J . W . , "An Investigation of Pore Water i n Compacted Cohesive Soils. " T . M . 654, U.S. Dept. of the Interior, Bur. of Reclamation, Denver, Colorado (1956).

7. Holtz, W. G. , and Gibbs, H . J . , "Engineering Properties of Expansive Clays." A . S . C . E . Transactions, Vol . 21 (1956).

8. Kruyt, H . R . , "Colloid Science I , I rreversible Systems." Elsevier Pub. Co., New York (1952).

9. Ladd, C .C . , "Swelling of Compacted Clay ." M.S. Thesis, Mass. Inst, of Technology (1957).

26

10. Lambe, T . W . , "Structure of Inorganic So i l . " ASCE Separate No. 315 (1953). 11. Lambe, T . W . , "The Structure of Compacted Clay." and "The Engineering Be­

havior of Compacted Clay ." ASCE Jour. , Vo l . 84, No. SM2 (1958). 12. Low, P . F . , and Deming, J . M . , "Movement and Equilibrium of Water in Hetero­

geneous Systems with Special Reference to Soi ls ." Soil Science, Vol . 75, No. 3, p . 187 (1953).

13. MacEwan, D . M . C . , Nature. Vol . 162, p. 935 (1948). 14. MacKenzie, R. C. , "Some Notes on the Hydration of Montmori l loni te ." Clay

Minerals Bulletin, Vo l . I , p. 115 (1950). 15. Norr ish , K . , "The Swelling of Montmori l loni te ." Faraday Society Discussion,

No. 18 (1954). 16. Pacey, J . G . , J r . , "The Structure of Compacted Soi ls ." M.S. Thesis, Mass.

Inst, of Technology (1956). 17. Prutton, C . F . , and Maron, S .H. , "Fundamental Principles of Physical Chemistry."

Macmillan Co. (1950). 18. Seed, H . B . , Lecture presented at M . I . T . September summer program on The

Design and Construction of Earth Embankments (1958). 19. , "Soil Mechanics fo r Road Engineers." Road Research Laboratory,

Her Majesty's Stationery Office, London (1954). 20. Taylor, D . W . , "Fundamentals of Soil Mechanics." John Wiley and Sons (1948). 21. Terzaghi, K . , and Peck, R . B . , "Soil Mechanics in Engineering Pract ice ." John

Wiley and Sons (1948). 22. Warkentin, B . P . , Bolt, G . H . , and M i l l e r , R . D . , "Swelling Pressure of Montmo­

r i l l o n i t e . " Soil Science P r o c , Vol . 21, p . 495 (1957).